Contelog: A declarative language for modeling and reasoning with contextual knowledge

Abstract

Context-awareness is at the core of many modern-day applications in safety and secure-critical domains. In existing context-aware systems knowledge and context are not formally integrated, and consequently adaptation behaviors for safety-criticality cannot be formally reasoned. In modern day smart systems, such as healthcare and advanced manufacturing, context-awareness must be combined with contextual reasoning in order that new knowledge can be inferred and based on which strategic decisions can be made. Consequently, a rigorous approach is essential to represent contextual domain knowledge and inference rules in order to combine the logic of domain-based decision rules with contextual constraints for contextual reasoning, and decision making. In this paper we address this later issue and introduce a formal approach to achieve contextual reasoning. The framework that we create, called Contelog, conservatively extends the syntax and semantics of Datalog to reason with contextual knowledge. In this setting, contextual knowledge is reusable on its own. The significance is that by fixing the contextual knowledge, goal-specific analysis rules may be changed, and vice versa. By providing a theory of context, independent of the logic of the rules, we have developed a simple and sound context calculus using which it is possible to export knowledge reasoned in one context to another context. Query processing and implementation of Contelogprograms convince us that it has the capabilities to reason in systems where perception and cognition are formally combined for problem solving.